Interview with Romain Bouqueau – CEO of Motion Spell on the ROI of Open Source Software vs. Buy vs. Build

Editor’s Note: In this interview, we talk to Romain Bouqueau, the CEO and Founder of Motion Spell, an open source evangelist and Video Streaming & Broadcast Technology entrepreneur about this thoughts on open source software, drivers of innovation, collaboration & globalization, the ROI of Build vs. Buy. vs Open Source and other topics in this interview.

Video Transcript starts here

Introductions

Krishna: Hi and welcome to the OTTVerse podcast. Each week we try and bring you the latest in video streaming technology, platforms and services. This week we have with us Romain Bouqueau, the founder and CEO of Motion Spell,  popular for their GPAC software, which I think is used in almost every media company for packaging and several other media tasks. 

Krishna: They were recently in the news for their integration into Netflix’s tech stack, and Romain has spoken about it in multiple places. He is also known to be a very strong advocate for open source software and is one of those few people who have gone on to make a business and has shown that you can actually do open source, make a living out of it, actually build a business out of it. 

Krishna: So we’d like to pick his brain on these topics today. Thank you, Romain, for being on our podcast. It was great meeting you in person at NAB and after many years, actually. And thank you for accepting this invitation. 

Romain: Hi Krishna. Thanks to you. I’m so happy to be here.

Krishna: Let me invite Romain to introduce himself, introduce us to Motion Spell GPAC and then we’ll take it from there. So over to you, Romain.

Romain: Yes. My name is Romain. I’m a developer. I’m a developer at core and used to be a musician and I wanted to do music, but I ended up into video just by chance and I’ve been in video for the last 15 years. And same for open source. I have contributed to projects, but that’s because, like so many people, I was frustrated by some things that I decided to contribute to several projects. GPAC was one of them. And yeah, eventually I got a job in a university that involved working on GPAC most of the time. 

And then I made this company that you mentioned, Motion Spell. And so Motion Spell is just it was a way for me to provide commercial services and licensing and everything that companies needed around GPAC. And it’s been a ten year journey to improve GPAC from the commercial perspective up to this point where we could sign with Netflix. And now things are crazy. It gives us a lot of visibility. 

Open Source being a pathway to Innovation

Krishna: I think that’s fantastic and congratulations on this journey.  This is why it is important to have you talk to us, because you’ve been in the thick of it all, made all the critical decisions. So let me just pick off on that thought. You felt that the open source world is an interesting world where you would like to contribute. 

Probably you found a lot of like-minded people. Why is it that we see a lot of innovation? Or why do we think why is there this popular notion that open source is a pathway to innovation? Do you have any thoughts on this? 

Romain:  Yeah, I think that there is the collaboration side. Humans are social beings, and so when there are difficult projects, we like to find people who think like us, want to try like us. I think it has never been so easy to find people all around the world and share your code and show it to other people and advertise it. 

It’s so easy now. And most of the case, it’s easier to go to a project and say, hey, but I would have done differently. And criticizing something that exists is easier for most people than starting something from scratch. You know, you see something and you say it’s not fast enough. I would have done this differently. It needs to run on the GPU. I want to have higher resolution or whatever. And that’s how you start contributing. 

There is a pain point, there is something you don’t like. You want to contribute, you start sometimes that’s the way the stories start. And then when it comes to R&D and I think there’s something really if you look at, Motion Spell is quite an innovative company. As I said, I used to work in a university. There is a lot of people in GPAC also still working in universities, a lot of open source software actually starting in universities at least a couple of years back. 

I think there is a ground for innovation there. And the problem is, if you look at the topology of customers of Motion Spell right now we have really big companies and really small companies. And I have the feeling that R&D is done by these really big companies pushing for new stuff because they have the capital and they have the humans and they can do it. But also the tiny individuals who carry a strong vision about things and they want to start something, I think then they need to meet at some point. And that’s the really difficult part. You can have really interesting ideas, but you need also, and that’s completely different skill set, you need to gather people around you. 

You need people to know about what you’re doing, why it’s important, and you need to carry your message and you need to carry it for a long time. When you’re a tiny entrepreneur and you’re isolated, you need to carry on. 

Big Companies working with Open Source Innovators

Krishna: Suppose I’m an excellent C programmer and I want to write the fastest JPEG 2000 decoder that has ever been written. But that would still require me to have access to the spec. And a spec requires substantial work from individuals of very large companies dedicating their time and effort. So it almost seems like there is this confluence of very large companies who have the time and effort to spare plus the sense of making stuff open or at least at a lower cost, that at least I can find the spec somewhere or I can pay for the spec. So how do you see this? It’s almost like they feed off of each other. 

Romain:  Yeah. So first I think being an excellent developer is not sufficient. You need to be an excellent developer, but you need to be something else, like being able to read specs, for example. And then I think you mentioned something that we’re going to try to cover after the cost of a spec for someone working in the US. Is not a problem. But if you are a student or if you work in another country or whatever, it can be a lot of money. 

And especially in our industry where ISO is the only international standard body and organization that’s allowed to publish these international standards. And the broadcasters have to use international standards or at least the use to. There is a legacy here and we are using dozens of standards. So if it’s $100 or $300 and you have to buy dozens of them, then that’s a problem. But the pirates, they don’t care. When you’re passionate about something, you find solutions, you get access to it, you find a way to hack them. 

The specs are not really a problem, but sometimes hardware is, because you can copy specs and you can find people. You can always LinkedIn someone and say, hey, I’m young and passionate. I would like to get access to this spec. And someone gets you a copy. But if you need a computer, then that’s another story. 

Krishna: It is. And it’s quite interesting that the software side of very large industries, which require R and D labs, I think those are also ripe for innovation. I can’t imagine myself doing cancer research sitting in my bedroom. But I could write a software that does image processing on cancer scans, CT Scans, and stuff like that. Right, so you do need data. 

Romain:  I agree. You do need access to specs. So those are  key drivers for innovation. 

Krishna: Anything else that you want to touch upon this as being a driver for innovation and why open source is important? 

Romain:  Yeah, I think that humans thrive when they cooperate. And I think that on this Netflix stuff, a lot of people said, why would they use open source for something as critical to their workflow as packaging? If you’re not using the best packager, then you’re going to waste bytes. So that’s a lot of bandwidth. You have 200 million subscribers when you’re on Netflix, so it has a cost for you. So you need to use the best in class. 

And when you’re on Netflix, the best way to do it is to do it yourself. And that’s what they did. But eventually they said, okay, that’s not where we want to compete. We want an open world. We want things that are interoperable. 

So they decided to go to open standards. And we were collaborating in MPEG already. And that’s one of the reason why they chose us. The differentiator of GPAC as an open source software is that we do a lot of R and D. 

I think that’s really strong for us. So this one and the second one is the existence of Motion Spell and all the commercial stuff. If you’re a company, you know, you can talk to reasonable people and find agreements and you know that you’re safe in this area. But there is a really, there is an increase in the complexity of stuff also. And I think that for companies being based in the US. You can have talented people, okay, but it costs you a lot of money to follow up with all of this. 

There is a Netflix engineer, Cyril Concolato, who went on stage with me at the NAB streaming Summit, and that’s what he told us. That’s impossible. That’s dozens of standards, and it’s not the work of only one person. At some point, like following all the standards and implementing the standards, doing it the correct way, dealing with interoperability, at some point you want to say, okay, I’m using an open source software. 

There are hundreds of users. Most of the interoperability is going to be solved by the other users, and I’m going to benefit from it. And when you are an innovative company like Netflix, you contribute on the standards. You can sponsor the addition of some features in GPAC for innovating on your end, but then you can rely also in term of cost and efforts on the other developers. Some people said, wow, they changed to this new packager and now they have that and they are doing Live. And I think people start to understand that if there is, I don’t know, VR in GPAC, and Netflix wants to go VR, then they can do VR in a Snap. That’s free for them now. 

Krishna: Yeah, so I’ll come to that part next. I think that’s something very important for us to talk about. Like, why would a company even choose? And if we can dive into that topic, it would actually enable a developer to also understand how to approach companies or approach problems, but to just wind up this last thing. 

It is a very large topic, but how does someone actually make a living doing open source? There are people who do a day job and night job. They hack some code. But if I really want to build this as a business, do you have some tips? 

Building a Business with Open Source

Romain:  Yeah, that’s a large topic to talk personally. You are the CEO of your life. Okay? So think about this. People say, I have a private life and I have a public life and whatever. When you’re passionate about something, there’s no distinction. You’re doing it because you like it. If you’re not doing it, sometimes you are in a position in life where you cannot do it. 

Think about what would improve your position for doing this. Being an open source doesn’t change anything. It’s not better or not better. You interact with a lot of people with good and bad sites, to be honest. So it’s not open source when it comes to open source. More precisely, I think, as I said, you need to be passionate. You need to feel like you’re solving a pain point for you at the beginning. And if you want to make it a living, it must be a pain point for other people, a sufficient pain point so that they decide to pay for it. 

And there is a transition that is sometimes when you look at videos on entrepreneurs, they tell. The person starting a company, leading the company from one person to ten person, is not necessarily the same one as the one leading from ten to 100 or to 100,000. 

You don’t need the same skill set. Some people can adapt and learn. And the Mark Zuckerberg or Bill Gates or Steve Jobs, they were tremendously be gifted with and probably well surrounded also by teams around them. But they could do it. Some other people, it will be really difficult for them. So they need to accept at some point that it’s not only their baby, there are other contributors and they need to shift to something that serves other interests. 

You need money, definitely. You need money for your life. You will need money for your project, a project like GPAC. As I said, we have R& D to run. We need around half a million dollars. Considerable amount of money. We need to go to MPEG, we need to buy the specs, we need to buy computers, we need to run the validation, we need to do everything. And it requires people to work permanently on the project. People don’t realize the scale of this. That’s the reason why ten years ago I said, okay, I don’t know if you know the bus factor. The bus factor is the number of people that needs to get hit by a bus for your project to disappear. 

Romain:  That’s pretty harsh. When you begin, you’re a genius, right? There is one person being able to be a superman on the topic of your project, on the code, on understanding everything. But at some point, if you want to scale, you will have average people on your project. 

Necessarily not superpowered people all the time. That doesn’t work this way, or you decide to do it like Netflix did, they decided to hire the best talents only, and they pay us a lot of money. But that’s a strategy. But when you’re open source, you have to see the people coming to you and work with them. And then there is this shift where you go from, I’m passionate about it. I’m doing it on my free time. That’s my passion to I do it as a living, and I accept that people pay me to do things that I wouldn’t have done. 

But okay, that’s also life. That’s their needs. That’s the way they use it. And probably it will drive you to other places, but you need to be passionate in life. As I said, for me, there is no distinction between the private and the public part. It’s a whole right. 

Open Source vs. Build vs. Buy

Krishna: So let’s switch our thinking hats a bit. And I want to put you in the shoes of a CTO or a CPO of a large media company much like Netflix. Let’s say that you have to replace your packager – something that you’re super familiar with. Having been a product manager myself, you’re always faced with this conundrum, right? Do I buy commercial software? Do I build? Can I assemble a team to actually build this from scratch? Or there is an interesting third alternative, which is picking up open source. 

I have never seen a clear cut answer to this problem. There is no magic ball that says, hey, take this, take this, take this. It’s always a gray area where sometimes even egos play a role in making the decision. Can you walk us through this framework? Can you walk us through, let’s say, build versus buy, commercial versus open source? And what’s the ROI out of all of this? 

Romain:  Krishna. We are going to write an article, I think. So this build versus buy versus open source. So, you know, it’s time, skills and capital when you want to build something and yeah, skills, that education. And I mentioned this because there is a talent shortage. 

I think in general, in tech and in the media industry, what you’re doing is really great because you educate and your publishing is more technical than most of the other stuff that just focuses on the business side. I think it’s important to highlight also the great projects, doing great technical things, and also because we face a really big challenge for the future, which is environmental sustainability. So I think that’s out of scope. 

Maybe for today, but again, for writing about this, you’re going to write a lot. I’m pretty sure about this in the future. So when you consider a solution, for example, we’re a specialist in packaging, so we sell a lot of packagers. So the first course that people see is licensing. So open source is free. It’s free. So, of course, if you’re successful, then it’s good to give back to the community. But at least you try it for free. 

You can go to the forums, you can ask your questions. That’s free. But if you had to pay for a license, that’s typically $150 per year and per channel, basically $150. Okay. Then there is a hidden cost in all the cases for packaging. You need to prepackage if you want to package live or package on the fly. You need to prepackage your content. So whatever you’re using, you need to prepackage it. So that’s integration, right? You need to spend the time to qualify your content and then you need to also integrate the software that you’re going to use, either open source or doing it yourself or buying it into your own workflow. 

So that’s the integration cost. And then of course, you need to run the software. You’re going to run it on a computer. So you have to pay for the computer and you have to pay for the bandwidth also because you’re streaming, so you have a lot of cost. So basically the server cost is not a lot. That’s probably $100 per year per channel. So you have 150 for the license or free if you’re open source or digital, then you have to pay for the channel, then you have to pay for the bandwidth, which is basically a couple of cents per gigabyte. 

So that can be pretty low. It depends because depending on the country where you are and the infrastructure, it can be more expensive or less expensive. You need to optimize it, right? That when you run and when you have viewers, distribution is your main cost. 

You need to optimize it. So at the beginning, if you don’t have many viewers, it’s not a problem if you don’t have the best packager or you’re not using the best settings for whatever reason because it’s always a choice. But then you need to move on. You need to find the time, as I said, for the integration. It’s like if you want to buy a vendor, you need to make a proper evaluation. So you’re going to spend days to write a request for proposal and to evaluate with the solutions because of course they are all going to say that they do it better than the other one. 

But when it’s not in the open. You need to compare it yourself. How do you compare it? It’s not always easy. So you could spend a couple of days, we could say ten days for this. If you go into open source, if you use GPAC, for example, it’s going to be a bit higher than that because a GPAC is a kind of raw material. It doesn’t maybe answer your exact need like a vendor would do, even though with Motion Spell, we would like to go this way and have off-the-shelf products. That would be another additional stream of revenues for us. 

But at the moment you need to spend more time. For example, you have the packager, but we don’t provide the Http stack. So when you get an HTTP request on your web server, then you need to read your own configuration and call GPAC, get the segment and serve it back. So it’s not complex. Like usually customers say that they spend ten days, ten more days on this. So that would be ten days, for example, for using a traditional commercial vendor and 20 days for using it with open source. 

And then of course, you have running cost like maintenance. It’s not because it’s open source or non open source that you don’t have to take care of your system and you have problems, you have too many people on the infrastructure. You know how it is. And what is interesting for me is to try to compare it, this structure of cost, for example, in different areas of the world, because you have substantial differences in how many days of work you need. 

And the cost of labor is completely different in different parts of the world. So imagine you build a packager. It takes you roughly one person per year, like 200 days per year. So imagine you have 200 days per year and you pay the 100. US dollars per year per channel for your server. Imagine then if you decide to go commercial, then you only have a couple of dozens of days and then you’re going to pay for the license. So there is the 100 plus plus 150 for the license, so that’s $250 per channel per year. 

If you go open source, then you have less days than doing it yourself, but you have more days than the commercial because we don’t provide a Polish product. So you have to spend the 20 more days on this. 

We’re at 50 days and you only pay for the server, so the $100 per channel per year. And now I made the math. Imagine you are a typical tier two broadcaster. You’ve been broadcasting for years. You have a catalog of a lot of content, 100,000 contents in your catalog and then you made your computation. You know how your CDN is going to cache your contents and you know that you would have to serve and to package on the fly or 500 concurrent content at the same time. So you have 100,000 in your catalog, but you only need to be able to package 500 at the same time. 

I hope you can share the figures with the people, because talking figures live is always difficult. Doing it Yourself if you’re in the US, you’re in the Silicon Valley and you pay a lot of money for your engineer, you have top notch engineer, you pay them $3,000 per day, right? 

All included that’s a lot of money. If you do this doing it yourself during a year, it already cost you $650,000 just to build it yourself. If you be the other extreme, imagine you’re in a country where life costs you nothing like 100 dollar per month. So that’s really the two extremes, right? If you do it yourself, it only cost you $70,000. So ten times less to do your own stuff compared to doing it in the Silicon Valley. And if you decide to go commercial, it’s going to cost you in the Silicon Valley roughly the same than using open source. 

It’s going to cost you roughly $200,000 to serve this use case. But if you go in an area yes, you have a question. Very interesting point. Please continue. I think I have something to add at the end of this. Yeah, it’s three times less expensive to go to commercial or to go to open source if you’re in the Silicon Valley compared to doing it yourself. So it explains why in the Silicon Valley, people just buy stuff if there is not a real differentiator like R&D, for example, and our standardization activity on open source for GPAC, a company like Netflix would just buy a commercial product, typically. 

But if you go to a country that would be where the labor is really not expensive, then open source and building are roughly the same. So you would have to choose, do I want to do it myself and have my own IP, or do you want to rely on something that’s already existing that’s almost the same cost? And if you want to buy something, it’s going to be two or three times more expensive. So that’s not an option to buy it. It’s not competitive. There should be different price listed depending on the labor cost of countries. 

But we live in globalized world, and so it’s really more expensive. In a country where the labor is less expensive, it’s really more expensive to consider buying a commercial product compared to doing it yourself, either from scratch or from open source. 

Communities in Open Source

Krishna: This is a fantastic perspective. I mean, it kind of changes your thought process on where open source would work and where commercials would work. I’d like to also get a different angle. I think the size of the company, the maturity, and the stage at which they’re in also matters probably in this decision making process. 

If I am trying to launch a new service, does it just make sense for me to buy something commercial, test my MVP? If I’m making money, maybe as a cost optimization, I’ll try and build it in house or bring in open source? I have not validated my business model yet. So does it even make sense? Plus something which you’re very familiar with, I guess, is labor and getting people to work on your project. And the time that a company has to make mistakes, which is very costly. 

I’ve seen projects go so horribly wrong just because people are very enthusiastic. It’s not something that they it’s just out of goodness of their heart, like, hey, let’s just build this. I’m an engineer. I know how to build. But then you don’t realize that either somebody in the open source world who’s been doing it for five years or a company has actually bled. People have cried, they have rewritten their software. They are where they are after probably two or three iterations. So do you have that time? Have you seen any examples like this in your experience? 

Yeah, there are so many things to say on, but even though the world is globalized, communities are not okay. They are vibrant communities everywhere in the world. They are vibrant, open source communities in India, in China, in the Western world. I think we are more unified. Unified for years. But you look at that. 

FFmpeg was starting in France by French people. Media info was targeted in France. VLC was targeted in France. GPAC was targeted in France. TS duck was starting in France. You cannot say that it’s a globalized community connected by the Internet. 

There was at the moment, these people didn’t know each other. There was a pool of knowledge and that’s all around. And that’s why education is so important. What we’re doing right now, we want the best of the best, the best communities. I don’t want a hundred. That’s what you said. I don’t want 100 teams working on the same project. Maybe we only need two, three, five teams taking different angles and competing on stuff. You don’t need 100 competitors. 

This is what’s happening in our industry right now with the content. It’s pretty sure that everything’s going to concentrate. And that’s the same for software, that’s the same for sustainability. What sense does it make to have so many people working on the same subjects? 

There is a talent shortage anyway, so let’s try to focus on this. About what you said, that’s the second time you mentioned it. Engineers, they want to push innovation, but we are not Apple. You cannot create a product and say people are going to buy it. 

It’s there at another level of marketing. Right? We’re not there. So at GPAC we do a lot of R & D and just for people to make sure to understand. So we choose what we want to do because we have limited resources. We are a small team nonetheless. And most of what we do just fail. It fails. We have decided to go on. It never really took up. So we had partners. We were spending money, and I was still use it, and it’s still here documented, et cetera. But I don’t make any business in it. And so when we discuss priorities, I have people like you said, who are passionate about it, who said, yeah, but I hadn’t finished. I’m pretty sure if I add this feature, people will love it and whatever, but just beyond our scale, technology is fascinating. 

Open Source Projects coming out of France

Krishna: So just going back to something you mentioned, what’s so special that led to so many important and impactful projects coming out of France? 

Romain:  Education. There were so many public researchers and students starting projects. People are not smarter in France than in any other country. So there were a pool of companies doing video. There were researchers working on video at the same time. And it created something where there were so many experts. 

If you go in the Silicon Valley, you can see it. There are so many people from France in many companies, it’s not a visible minority, right? But when I discuss with people, I said, but you’re French, right? You hear the accent of the people. Many of these people went there. My competitor, Bento4  was started by French Guy Or. So it’s really weird to see that there are so many founders of Open Source Project who were actually in the same situation at the the same place at the same time. 

And that’s what I hope for any disc community stuff just creates an industry in a country. It makes things really easier. We are really inclusive in GPAC when it comes to contributors. But as I said, the world is globalized. But there is also a sense of culture that makes it easier, I think, for people of the same culture to work together naturally. Okay, again, that’s education. I’ve been working with people from all over the world. I understand how I try at least to understand how people work because we are all different. And open source doesn’t escape from this. When you build a team in a company, if you have people from different countries, there will be these weird or funny moments where there is some misunderstanding. 

Challenges in Open Source Software

Krishna: Absolutely. I think one of the craziest, the most recurring problem I’ve seen on mailing list is top posting versus bottom posting. Somebody top post a question and everybody like the rules of this community is to bottom post and somebody bottom post and they say so yeah, you have a lot of problems which are not even technical in nature. Right? And they can tick people off from across the world. But it’s also interesting how it’s like a self regulating society. One of my favorite projects, I would say is Debian. Whenever I get a spare computer, I just wipe it clean, put Debian on it. And I’m so appreciative of the fact that you have thousands of contributors across the world just working towards a single minded common goal. Fantastic stuff. I like it. Yeah, that’s also one of the big open source project. 

Romain:  I’m talking to you with Linux from Firefox. And recently I deployed a DBN on my car just because I got crazy with the infotainment stuff that he put on it. And actually I took the open source version of Android. I found people having customized it to a certain level. So I was plugged on right before we were talking. I was plugged on my car because the firmware has bugs. That’s the magic of open source, right? 

You’re fed up with something. My car just drives me crazy. There are so many bugs. I can, at least in some days, my own music. It’s just stuck in the middle. I don’t understand why, but the where people having the same pain point, and we’re discussing on the forums, and we’re trying to find solutions and trying on our car how to modify it. 

And Debian is a really important part of it because they have, as I said, a vibrant community. I think that’s so important. If you want to contribute to an open source project or if you want to choose an open source project for your business, look at some metrics. How many are people active answering questions? If it’s on GitHub, does it have stars? Do you have issues that are regularly closed, or are there like hundreds or thousands of open issues? I think sometimes you’re passionate about something. You don’t look at this, but you need to look at these points before choosing your project. And Debian definitely a vibrant community. You can get answers to your questions. It’s really active. I’m loving it too. 

Krishna: Yeah. So actually, that point that you mentioned turned out to be very critical. In my previous roles, where we actually wanted to find a user agent password for one of our modules, we looked at a bunch of GitHub projects, and I think this was one of the big things that we highlighted. If a question is asked, is it answered in a reasonable amount of time? When was the last check into that project? Was it a month ago? Was it a couple of years ago? The major authors, we did a bit of research on them. Are they still contributing? I think. This also adds a bit of pressure right on the open source community, where you’re judged by certain factors when you’re not receiving probably enough. So that brings me back to the other side of what do you think companies should also do to encourage open source? 

Romain:  Because they might not have the time to do R&D. There are mid sized companies, small companies, or individuals. What do you think companies can do in this regards? I think it’s been ten years, eleven years now that I’ve created Motion Spell and I wanted this to foster open source for GPI. 

But also in other companies, there have been so much progress. Companies ten years ago, many companies didn’t want to hear about open source at all. That was a virus, okay? They didn’t want to touch it. But now they accept it. They accept that they can contribute. They can create their own project and open source them. Things evolve. I think it’s again, everything is education. We need to understand each other. 

We need to understand the benefits of doing so. There is no reason to keep it closed. When you’re dealing with a problem on a closed source software and you have no solution, you would be so happy to have something that is open and just dedicated the part of one of your engineer to help to solve this problem or solve another problem. 

Think about this. Sometimes at some point, we need competition. There are really hard problems that we are trying to solve. And we need competing companies, is with capital or whatever. But for most of the people of our lives, it’s already solved. It’s already solved. We don’t need to compete anymore. What we need is what Netflix did. We need open standards. We need open source software. And okay, there can be competition between commercial and open source and whatever. 

But my feeling is that at the point where we are, we are going to take all the market because at some point there will be so many people using us that this is going to be critical infrastructure and then people will say, okay, there is no real option. And maybe at some point, like any company, we’re going to stay too quiet and a disruptive innovation is going to come and kick us out. But I’m going out of your initial question. I don’t know exactly. I think it’s a question of education and leadership and that can go from the bottom or from the top, but the company needs to agree to share and to contribute to project. Right? I think this is there in a couple of different avatars. 

Krishna: But  what I really like is the Netflix Prize, which you probably know about, which was more than 10-15 years ago, probably ten years ago, where they had a price for writing personalization and recommendation engine software. It was called a million dollar price or something. Do you think something like this companies could sponsor where they say I need a better way to, let’s say package with DRM on top of mp4box. Here’s the open source code, here’s this and we will contribute it back but the winning contribution gets $25,000, something like that. 

Is that the way people should be also thinking about this? Because they need to be right. That’s what they do. 

Romain:  That’s what they do. People come and pay for specific features and sometimes experiments. And as I said, my main problem is the talent shortage. I don’t have any skilled. People to execute. So I need to focus on what is the core for us, what is the most interesting for us, what keeps my talent focused on the project and doing things. That’s my really main problem. But when people come and say, I would like to pay you to do this, that’s our business. So we try to say yes whenever we can with everything we can do. And I think that’s a way for us. 

You were talking about how to go from open source to being a professional open source developer or creating a company. That’s exactly how it works. People pay us for things that they need, so it drives us in the right direction. 

Krishna: That’s really positive. That’s a positive loop. Okay, so if I play devil’s advocate now to you and you say, okay, Roman has built XYZ packager, it does HLS, it does Dash, you pay me to integrate your DRM vendor, but then Krishna: comes along and says I’ll open source that part as well. So you see this conflict which is taking place. How do you know when to stop open sourcing stuff? If that doesn’t make a bunch of people angry? I’m just going to ask that question. 

Romain:  Yeah, that’s a real question. That’s a real question. Because on the commercial side, we are doing things that are against freedom. Like DRMs. Yeah. And everyone I know has an ad blocker. I’m using Linux, and the main focus of the industry right now is to put ads in everything. So, of course there are real issues with this. And as you said, it doesn’t need to be fully open source, like free software to proper. I think there is a scale in it like people can pay for something and it can become open source only a couple of months or a year later. 

Also that when they are competing implementations. Because I’m licensing the software I need to take care of the IP. So I have quite some control with the guys I know and I trust. And the core of GPAC we are, we are a huge majority of GPAC. So nobody is going to fork the open source software, I think, to do things that could happen, that may happen, that has happened to FFmpeg for example. But this is not the case for us. I think we have a strong leadership. And then there is the other thing we mentioned and we took figures on the, for example, the just in time packager. What is the moment where we decide to do it ourselves and to sell it? What are we going to do? 

Are we going to open source it completely? Are we going to give the source code with a proper retail license? But I like to give the source away because it lowers the support for us. People can recompile it if they need to update their operating system. They don’t need us. They don’t need us to reproduce and do whatever. They can take their own stuff and they can rebuild it. And if they need for example, 24/7 support that we don’t supply because we’re a tiny company, they can have their own team and they can find a quick patch and they can revert on the versions or whatever. 

They don’t rely on us. So I like giving the source code away. There are moments where we don’t give the source code. There are moment where we give the source code but it’s not open source. There are moments where we delay the moment where it’s going to be open source and there is the full open source core which is 99 9% of the code. 

At the moment of speaking, but some companies like, I don’t know, Netflix could come to us and say, I want you to integrate this, but this is going to be open source in a year to just pay us for this. That’s a possibility. That’s an interesting thought, but again, I see your answer actually going back to the big word, talent. Right? If you can have a BMW or a Mercedes, but if you don’t want to drive, it’s of no use. 

Krishna: Even if you give away your code for free, someone should have the talent, I think, the time and the patience to actually read the code, understand how to modify it. I don’t know. Many companies would be willing to take that bet, honestly. It’s a little risky to it’s like changing the wheels on a car that’s running, right? I don’t know how many people would take that risk. So, yeah, that’s something that we heavily discussed five years ago, because that’s not a news topic. 

Romain:  That’s a topic that arises from time to time. And GPAC, we improved dramatically. The README, the wiki. We have hundreds of pages, some of them written by humans, some of them automatically generated from the code. We have a lot of tests. We have a test suit with hundreds, if not thousands of tests. We’re just feeling the confidence that if you want to contribute, if you do something wrong, you’re going to know it immediately. 

So you’re going to gain confidence that when you propose a patch, it’s not going to be rejected. You’re not going to break anything else just because you have this test. You have this documentation, you could read it, you could actually communicate with the maintainers of the software. 

So trying to create these environments of trust. And yeah, kindness also we try to be kind. Sometimes users are pretty difficult for us. They are really rough on us or whatever. But we try to stay calm and answer the question. Sometimes we have to ban people and it happens and sometimes people are really weird. We had contributors on the issues, on issues on our tracker recently and we were solving real difficult problems. 

They were mostly asking questions because they had problems and they could see some relation when packaging with MP4 box or whatever. Once the problem was solved, they deleted all the issues. So yeah, that’s you know, this is like this is really like in life, right? 

There is a life in the open source. There is a life in this community where some people behave correctly, some people don’t. You have to take the action. And when it comes to a company again, that’s education and that’s a spirit. If you’re ready to do things, choose wisely your project and when you contribute, okay, this won’t work the first time, but you’ll get there. And GPAC is not the only project who have test and a Wiki and Kind community. There are tons of projects like that. I would say even that most of the projects are just led by kind people willing to solve your problems and help you. Yeah, big people need to be confident and they need to go with the right spirit. 

Romain:  They want to contribute, but they don’t know how to do it. They are willing to work to contribute. It’s not like you go on a forum and you ask questions all the time and you never learn. You know that it’s complex subject, as you mentioned, which is something that can be scary. But you need also to. Educate yourself on these subjects to be able to contribute in a meaningful manner. 

Krishna: Okay, so, Romain, coming to the end of this discussion, I’d like you to help us summarize this entire free flowing discussion, coming back to our original premise. What is the ROI of open source? So if you could tell us the ROI from a developer’s perspective and an ROI, you from a company who’s trying to evaluate whether I do build buy versus open source, if you could just help us summarize, that would be very helpful. 

Romain:  Yeah. For developers, I think if you’re a talented developer, the ROI is really high because maybe the hours, like the hundreds of the thousands of hours that you’re going to spend in it, you’re going to educate on the subject. You’re going to be an absolute expert. People will see your work. And a lot of people I know got hired because of this, because they made for example, Hls.js was made by Guillaume du Pontavice. He used to work at a company doing his own stuff. 

And then being hired by DailyMotion and then being hired by Netflix. It just puts you on the rail to something really different. For companies, I think open source, it’s a bit different. I think you need to think about what role you want to have, because socially, for sustainability, for the image that you project of yourself, it can make a difference. I also believe that it makes a difference economically. So that’s the figures that we gave. I think open source is always a viable solution for you, but you’re competing between building yourself and buying commercial. 

You need to choose your battles. Right. You took the example of building a startup at some point and using off the shelf just to validate your business model. That’s exactly what we did. Like five, no, seven years ago, we made a company, and I remember, to validate the business model, we decided to go with Woza. 

As a media server, I could have use GPAC and integrate GPAC or whatever. It’s just that at the time, it was really convenient. Wowza proposed hosted instances for $100 per month, and that was easy for us. We said, okay, instead of spending days that would cost us hundreds of dollars already, we just pay for this, and for a month we can try if what we’re doing works or doesn’t work for the market that we target. So it’s always different. Each company is different, each individual is different, each open software is different. Absolutely. I think it’s that difference which is important, gives us different perspectives. 

Conclusion

Krishna: And I wish actually, more people would come out and talk about this. If you have anybody you think could come on the show, I’d be more than happy to talk about them. I think we need these different perspectives on how people approach the buy versus build versus using open source. It’s a fascinating space. I think the contribution of open source has revolutionized the streaming industry. I don’t know where a lot of companies would be without GPAC, without FFmpeg, without HLS.js. 

Lot of these things are heavily underestimated. I mean, I think most people, when they start a project, they just download a copy of all of this and put something together and push it out there. It’s hilarious, actually, that we’re even discussing this topic. It in today’s day and age, when many companies have made their fortunes off of these software. Even android, right? Even android ExoPlayer. You can go in. You can understand how everything works. Inspecting the code itself is an education in itself. 

I will, I think, take this opportunity to pause here, Romain: . This has been a fascinating, free flowing discussion. I think every time someone has you on their show, the amount of knowledge that comes out is great. 

Romain:  Thank you. Thank you for having me. Krishna, it’s always great to talk to you. Thank you.